This paper presents the history matching of a clean-up operation of a long horizontal oil well at the Oselvar field in the North Sea using a commercial multiphase flow simulator. Data recorded during the actual clean-up is presented in the paper, and is compared with simulation results.Well clean-up is the process of flowing the drilling and completion fluids out of a new well, removing formation damage and filling the well with formation fluids. The role of upfront simulation of well clean-up is to optimize the operation and provide the basis for the operational procedure to be used onsite. Also, after the well clean-up has been carried out, the simulation tool may be used for history matching in order to gain understanding of what happened during the operation and assess the quality of the upfront simulation model.The data from the Oselvar clean-up operation revealed that the liquid flow rates at the initial choke size were significantly different from the rates predicted by the upfront simulations. Also, the time it took until no more drilling mud arrived topside was longer than expected. The data shows that the heel of the well was producing for several hours before there was any production from the toe of the well. This is attributed to the high initial productivity of the heel of the well, and to unexpected rheological behavior of the drilling mud. Constructing a transient downhole boundary condition based on the recorded data and on interpretation of petrophysical well data, a history matched simulation model was built that gave good agreement with the data.The work in this paper contributes to improved understanding of mud retention in wells during clean-up operations. The data and simulation results demonstrate why the modeling approach widely used in the industry may lead to a conservative estimate of the pressure margin before a well is killed, and an optimistic estimate of the time that is required to clean the well.

No abstract

No abstract